def _get_write_kwargs(cls, fmt: Optional[FileFormat],
path: Path) -> Mapping[str, Any]:
t = cls.get_typing().io
real_suffix = CompressionFormat.strip_suffix(path).suffix
kwargs = t.write_kwargs.get(fmt, {})
kwargs.update(t.write_suffix_kwargs.get(real_suffix, {}))
if fmt is FileFormat.json:
# not perfect, but much better than the alternative of failing
# I don't see a better solution anyway
kwargs["force_ascii"] = False
elif (fmt is not None and fmt.supports_encoding
): # and IS NOT JSON -- it doesn't use "encoding="
encoding = kwargs.get("encoding", t.text_encoding)
kwargs["encoding"] = Utils.get_encoding(encoding)
return kwargs
def _get_read_kwargs(cls, fmt: Optional[FileFormat],
path: Path) -> Mapping[str, Any]:
t = cls.get_typing().io
real_suffix = CompressionFormat.strip_suffix(path).suffix
kwargs = t.read_kwargs.get(fmt, {})
kwargs.update(t.read_suffix_kwargs.get(real_suffix, {}))
if fmt in [
FileFormat.csv,
FileFormat.tsv,
FileFormat.properties,
FileFormat.lines,
FileFormat.flexwf,
FileFormat.fwf,
FileFormat.json,
]:
encoding = kwargs.get("encoding", t.text_encoding)
kwargs["encoding"] = Utils.get_encoding(encoding)
return kwargs
def is_text_encoding_utf(self) -> bool:
return Utils.get_encoding(
self._text_encoding) in ["utf-8", "utf-16", "utf-32"]
def test_encoding(self):
assert Utils.get_encoding("platform") == sys.getdefaultencoding()
assert "bom" not in Utils.get_encoding("utf8(bom)")
assert "bom" not in Utils.get_encoding("utf16(bom)")
assert Utils.get_encoding("UTF-8") == "utf8"
assert Utils.get_encoding("utf-16") == "utf16"